Reasoning on CURE-Bench (phase2)
83.803AccuracyTxAgent
Evaluation Results
| Method | Links | |
|---|---|---|
| TxAgentNumber of Executors=10, Analyst Agent=gemini-2.5-flash (with search), Number of Analysts=3, Architecture Configuration=Config B: Stratified Ensemble2025.12 | 83.803 | |
| TxAgentNumber of Executors=10, Analyst Agent=gemini-2.5-flash, Number of Analysts=3, Architecture Configuration=Config B: Stratified Ensemble2025.12 | 81.367 | |
| TxAgentNumber of Executors=30, Analyst Agent=gemini-2.5-flash, Number of Analysts=3, Architecture Configuration=Config A: Global Pooling2025.12 | 80.51 | |
| TxAgentNumber of Executors=30, Analyst Agent=gemini-2.5-flash, Number of Analysts=1, Architecture Configuration=Config A: Global Pooling2025.12 | 79.311 | |
| TxAgentNumber of Executors=10, Analyst Agent=gemini-2.5-flash, Number of Analysts=1, Architecture Configuration=Config A: Global Pooling2025.12 | 78.527 | |
| TxAgentNumber of Executors=1, Analyst Agent=gemini-2.5-flash, Number of Analysts=1, Architecture Configuration=Config A: Global Pooling2025.12 | 74.701 | |
| TxAgentNumber of Executors=30, Strategy=Self-Consistency2025.12 | 73.508 | |
| TxAgentNumber of Executors=10, Strategy=Self-Consistency2025.12 | 72.673 | |
| TxAgentNumber of Executors=12025.12 | 69.325 | |
| gemini-2.5-flashNumber of Executors=3, Strategy=Self-Consistency2025.12 | 65.678 | |
| gemini-2.5-flashNumber of Executors=12025.12 | 63.104 |